• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 99 subscribers
  • Views 740 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

HDR video?

Simon Gustafsson
Prodigy 90 points
Other Parts Discussed in Thread: DM385

Hi,

We are looking at the possibilities of encoding HDR video using an omap4460, and would like to know about possible limitations before we dive to deep into this subject.

Our sensor candidate can be set to use alternate row timings for HDR. (odd lines will have a different exposure time then even lines, and merging information from both lines could give HDR video, either at an increased bit depth, or after a process called tone mapping using “normal” bit depths.

1) Is this an application that TI have ever considered?

2) Does there exist any ducati codecs encoding at higher bit-depths then normal?

3) Would we really have to do the tone mapping in the main CPU? (Is there any way we could use the ISP? Is the hardware even able to look at non-adjacent lines and do more than simple filtering operations?)

4) Would there be some other way of encoding HDR video that we aren’t  aware of, or a different product. (and btw, the provided dynamic range extension is not an option for us)

As far as possible, we would like to avoid having to use either the CPU or an external FPGA to do the video processing.

BR
/Simon

  • 0
    Expert 1640 points

    Hi Simon,

    While there are existing TI solutions for utilizing an alternate row sensor for HDR purposes, a lot would depend on the final application & the overall system you are targeting too.

    I believe there is a DM385 IPNC reference design utilizing a similar sensor, so it's certainly possible to use a similar method on the OMAP4. In this case, the tonemapping can be done on the SIMCOP hardware accelerator set, leaving the main CPU untouched.

    I cannot comment on codecs with higher bit-depths, but as far as I know, if you're going to be configuring the ISP for YUV output - then you're constrained to 8-bit output. So whatever tonemapping that needs to be done has to be performed at the RAW stage and the unified frame fed back into the pipeline. This is assuming you need AWB and the other bells and whistles the ISP offers. 

    All this, of course would depend on what resolution, frame-rate, codec, etc you are aiming for.

    Regards,

    Akshay

  • 0
    TI__Mastermind 24525 points

    1) Is this an application that TI have ever considered?

    Yes, but not yet implemented on OMAP4 SW.

    2) Does there exist any ducati codecs encoding at higher bit-depths then normal?

    No

    3) Would we really have to do the tone mapping in the main CPU? (Is there any way we could use the ISP? Is the hardware even able to look at non-adjacent lines and do more than simple filtering operations?)

    You can use GBCE block of ISP to do tone mapping. It is not being used in video scenario. Currently SW is configured to use it only for still capture.

    4) Would there be some other way of encoding HDR video that we aren’t  aware of, or a different product. (and btw, the provided dynamic range extension is not an option for us)

    See Akshay's comment on DM385

  • 0 in reply to Akshay Panday
    Prodigy 90 points

    Thanks for the reply.

    I probably found what you mean. The SALDRE on DM365 User Guide suggests that you can process 720p @ 30Hz @ 58% cpu load (but that would be a 300MHz ARM926EJ-S, so I would expect a much lower load on an omap4, even at a slightly higher resolution (1280x960, same framerate)). There was no mentioning of using the IVA-HD blocks, so I assume that video encoding will be unaffected.

    I've found Appro's products, assumes those was the IPNC designs you mentioned. Will query them for more information about the sensors, if they use alternate row HDR (or if their solutions would be close to our needs anyway).